AN ANALYSIS OF NCAP SIDE IMPACT CRASH DATA Hansun Chan James R. Hackney Richard M.

Morgan National Highway Traffic Safety Administration Heather E. Smith ConradTechnologies,Inc. United States PaperNumber 9%Sl l-O- 12 ABSTRACT Since 1990, the National Highway Traffic Safety Administration (NHTSA) implemented a dynamic side impact compliance test. This compliance test, Federal Motor Vehicle Safety Standard (FMVSS) No. 214, is a nearly right angle side impact in which the striking vehicle moves at 53.6 kmph into the struck vehicle. In 1997, NHTSA begantesting passenger in side impact in the cars New Car Assessment Program(NC@). In the USA NCAP side impact, the striking vehicle is towed at a 8 kmph higher speedthan in the compliancetest. An analysis has begun on the data from the first NCAP side impact tests, thirty-two in number. In the crashes, accelerometers were installed in the door and door frames of the struck vehicle. Using the accelerometers on the vehicle structure and in the side impact dummy, the crash event was investigated. One tool used in the i:avestiga+ion the velocity-versus-timediagram. was First, the crush of the interior door and its relationship to the severity of the occupantinjury readings was examined. A correlationwas found betweenthe single independentvariable, amount of the interior door crushed by the occupant,and the Thoracic Trauma Index. Second, the datawas examinedto determinethe extent to which the pelvis of the dummy was loadedinitially beforeloading the torso. A weaker correlation was found betweenthe time duration (that the pelvis was loadedbefore the torso) and the Thoracic Trauma Index. Finally, the effect of the two independent variablestogetherwas examined. INTRODUCTION Based on the most harmful event, side impact accountsfor 25 percent of fatalities for passenger and car light truck crashesin the USA. [l] For passenger cars,side impact accounts for approximately 30 percent of the fatalities in passenger crashes. Likewise, side impact car accountsfor roughly 15 percent of light truck fatalities.

Interesting paper if we wanted to look at injury and crush

Since the use of dynamic Federal safety standardsin side protection began, in recent years occupant protection in side impact crasheshas receivedincreasinginterest. This interestcomesfrom both the consumers the automotive and industry. [2,3] In comparisonwith frontal collisions, the space betweenthe occupantsand the intruding element in side crashes is extremely small. In addition, the side impact crash occursmuch more rapidly. Consequently,occupant protectionin side crashespresentsa challengeto engineers designing a vehicle for safety. Significant researchwork, both theoretical and experimentalin nature, hasbeenperformedto characterize the safetyperformanceof vehicles in side crashes. Gabler et al. [4] analyzed data of 28 production vehicles that underwentside impact crashtesting. Theyfound that these vehicles varied dramatically in their ability to protect the occupantin the struck car. They were able to identity a design parameter- the door effective padding thickness (DEPTH) -that strongly correlatedwith occupanttboracic injury potential. Hobbs [5] investigated the influence of car structureand padding on side injuries. He analyzed more than 40 full scale vehicle impact tests. His findings revealed that a most important factor, in influencing protection, is the vertical intrusion profile of the incoming door. It appearedthat controlling the vertical intrusion profile of the door is much more important than the preventionof the intrusion itself. HobbssaysThe degree of door tilt has beenfound to influence the way loads are transferredto the occupant. When the door tilts in at the top, loads are concentratedon the thorax. Where it (door tilt) remainsupright, the loads are more evenly distributed and it may be that earlier loading of the pelvis reduces thoracic loading, by helping to acceleratethe occupant sideways.Along the sameline, Saabengineers developed a collapsebehaviorfor the B-pillar that reduces injury the readingsof the occupantsin a side impact. It allows the lower part of the B-pillar to behavemore softly than the

2490

upperpart. Saabfeels this collapsediverts crash loading to the parts of the body that can withstand them the most, the pelvic region. Saabfeels this collapseprotects the parts more susceptibleto trauma: the head, rib cage, and chest. [6] Lau et al. pointed out that the maximum velocity of the intruding door (of the stuck car) is important because the door strikes the occupantdirectly. [7] They compared the door motion to a powerful punch to the dummy. In s their paper,they pictured1 velocity of the intruding door the as rising as high (in magnitude) as the velocity of the striking barrier. Strotheret al. presented from another data crash that suggested velocity of the intruding door rose the to a lower level, roughly ,theterminal velocity of the struck Cal-.[S] Finite element modeling has been successfully and extensivelyused to simulate collision events. Using the finite elementprogram code DYNA3D, Rao et al. [9] simulateddifferent impa scenarios 45 kmph for a midct at size car being impacted in the side by a moving barrier. Their simulation results indicate that one can gain an understandingof how the interior door might respondto structural changesmade in the struck car. Blaisdell et al. [lo], in their comprehlensiveexamination of collision performanceof automotive door system , concludedthat ...merely increasing door and latch strength without consideringthe entire systemwill not necessarilyprovide additional occupant protection, and may be counterproductive.. . They recommend that designers graph the velocity changesfor different portions of the structurewith respectto the occupant. This recommended approachparallels the methodusedin Reference4.

TEST PROCEDURE

The vehicle impact tests that generatedthe data usedin this analysiswere conductedin accordance the with test procedureof the side impact NCAF The NCAP side . impact test is based on the dynamic requirements of FMVSS No. 214, but is conductedat a higher speed. The NCAP tests, which simulate an intersection collision, were conducted with a moving deformablebarrier (MDB), as the striking vehicle. The 1360kg MDB was moving at a speedof 61 kmph and at an angle of 27 degreesoff the perpendicularto impact a stationeryvehicle, as shown in Figure 1.

SPEED TRAP

ORWARD FTRAvEL j

Figure

1. Test Setup.

Since 1997,the NHTSA has carried out forty-six full scaleside impact tests under NCAP. Thesetests were conductedwith extensive instrumentationso as to provide data neededfor conductmgresearchaiming at improving vehicle side protection. Accelerometers were installed in various locations of the test vehicle including the door panels,A- and B-pillars, sills and floor, and vehicle center of gravity (CG). This information, combined with data recorded from occupants, usedin this studyto investigate is the differencesin safetyperformance and identify design parametersthat influence vehicle side crash protection. The velocity-versus-timeanalysisas previously referenced [4, 7, 8, lo] was used in this study. The authors feel it helps in the visualization of the kinematics of the occupant and the behavior of the intruding vehicle structure.

Each impact test used two side impact dummies (SIDs) as specified in FMVSS No. 214. One SID was positioned in the driver seat, and a second SID was positionedin the rear passengerseatbehind the driver, as shownin Figure 1. The dynamic response eachSID was of recorded accelerometers by installed at the upperrib, lower rib, lower spine, and pelvis of each SID, as shown in Figure 2.

2491

LATERAL SHOCK ABSORBER

LOWER RIB

I
A / REAR VIEW OF SIDE IMPACT DUMMY @ID)

LOWER SPINE

Figure 2. Rear and Side View of SID.

Twenty or so accelerometers were installed at variouslocationsof the test vehicleto m o n itorthe motion of the test vehicleand its structuralcomponents. Since the vehicle side doors and the door frames play an important role in side impact protection, special instrumenration usedto capture dynamicresponses was the 0 these components. For the front door, three accelerometers installedon the interior surface the were of

inner door panel. For the B-pillar, two accelerometers were mountedon the interior surface of the inner door panel. Shownin figure 3 arethe general locationsof these accelerometers. Actual locationsof theseaccelerometers mayvary with the individual test because the variations of in vehicledesign.Two accelerometers wereinstalled each in theA-pillar andB-pillar. Oneaccelerometer located was in the base the otherin the m id-section the B-pillar. and of

#5 #6 #8 #9 #12 #13

Left Side Sill @ Front Seat Left Front Door on Centerline Mid-rear on Left Front Door Left Front Door on Upper Centerline Left Lower B-Post Left Mid B-Post

Figure 3.

Location

of Accelerometers.

2492

PHYSICAL EVENT OF A SIDE CRASH AND THE VELOCITY-VERSUS -TIME DIAGRAM

Occupant t@ /

,- Interior Trim Panel ?

During a car-to-car side collision, the physical event is a complicatedtransfer of momentum from the striking car to the struck. car. To a largeextent,the severity of the crash event, as s.een the occupantin the struck by vehicle, is determined the time rate of changefor this by momentumtransfer. The time rateof momentumtransfer, in turn, is dependent upon the relative structuralstiffness and effectivemassdistribution, amongotherfactors,of the individually struck cars. Because their proximity to the of impacting car and the occupant,the doors(front and rear) and the pillars (essentially the A- and B-pillars) of the struck vehicle are among the componentsthat play a critical role in decidin,ghow the momentumtransfer is being carriedout aroundthe occupant. The doorsand the pillars use their energy-absorbing capability and their material strengthto channelthe momentum transfer. In addition, the intruding door structure can provide an interior surfacethat crnshesat a non-injuriouslevel and acts to protect the occupant. The characteristicsof the dynamic interaction betweenthese componentsand the vehicle occupants(the SID test dummies)determinethe effectiveness of the vehicle side crash protection performance. One useful tool, to understand the dynamic interaction betweenthe intruding door structureand the vehicle occupantsduring the impact, as well as to assess the efficiency of the door designin collision performance, is the velocity-versus,-timediagram. This diagram graphicallytracesthesecritical structuralcomponents and the responsesof the occupant. Shown in Figure 4 is a simplified illustration of a typical door construction. The door is generaIlycomprisedof the outer and inner panels (usuallymadeof sheetmetal) and the interior trim panel (usuallymadeof plastic:with or without energy-absorbing padding). Housedbetween the inner and outer panels (skins)arethe window mechanism, remoteactuatinglevers and rods, as well as reinforcing guard beam(s), if so equipped.The door is attached the doorframe,which is to comprisedof the pillar structure,roof rail, and door sill. The door frame is designedto resist collision forces and also servesto transmit crashloadsfrom the region around the occupant (essentially the doors) to other vehicle structuresduring the mash. In the NCAP side tests, the motions of the striking vehicle; the doors, pillars, and occupant of the struck car; and the struck car were electronicallymonitoredusing accelerometers.

f=T/;;

~~

/ / Striking Vehicle @- j ._-,.

Figure 4. An Illustration and Essential Locations Time Diagram.

of Door Cross Section of the Velocity-versus-

A velocity-versustime plot, typical of the NCAP side tests, is shownin Figure 5. The outerpanel (skin) is struck by the impactor(MDB) and movestogetherwith the MDB almost immediatelyafter contact, as shown by the curve denotedas 0. Within 3 to 5 milliseconds (msec), the velocity of the inner panel (togetherwith the interior trim panel) risesto the speedof the striking vehicle as it (thedoor)continuously undergoes deformation.Sometimes the speedof the inner door panel overshoots that of the impactor, as shown by curve 0. In fact, curve 0 is representativeof the velocities of the pillars as well. Dynamic contactbetweenthe door inner surfaceand the SID (in the areaof the ribs, spine, or pelvis) generally starts 10to 20 msecafter the impact eventbegan. After the dynamic loading of the occupant(SID) by the intruding door structurehas begun,the occupantusually reaches its peak velocity around 20 to 40 msec after initial impact. Curve @ shows a typicai velocity - time traceof the SID response. This response typifies the motion of the SID s rib, spine, and pelvis. As shown in the diagram, the occupantwas contactedand loadedby the intruding door structurestarting at time b The intruding doorcontinued in contactwith the occupant until the two separatedat time t, The continuousdeformationof the door structure can alsobe visualizedas a two-sequence event. First, the door starts to deform under the influence of the impactor. As this deformationcontinues,the interior door encroaches until striking the SID which resiststhe door motion with s its inertia force. This inertial loading of the door by the SID andthe impact loadingof the occupantstartsat b and lastsuntil time t, when the two separates. The velocity of the impactor, curve 0, and the velocity of the struck car, curve 0, generallymove to a common velocity, V~ The

2493

velocity-versus-timediagram not only documents(with a high degreeof clarity) the key interactions of the crash event but also supplies the necessarydata neededfor analyzingandassessing vehicle collision performance the s in a side crash. In the sectionsthat follow, the data in the velocity-versus-timediagram are used in an analysis and assessment the NCAP side impacts. of

Figure 6. Interior Thunderbird.

Door

Velocities

of

Ford

to

t1

TIME

Figure 5. Velocity-versus-Time

Diagram.

In analyzing all thirty-two NCAP side impact tests, the maximum speedof the interior door varied over a rangefrom a low of roughly 32 kmph to a high of about 59 kmph. For example,Figure 7 gives the velocity-versustime diagram for the 1997 Toyota Corolla. In the case shownby Figure 7, the maximum speed the door appears of to rise only to about the final velocity, vh of the two vehicles around 10 msec.

MAXIMUM

SPEED OF THE INTRUDING

DOOR

Figure 6 graphically presents the velocities of many essentialpoints during the NCAP side impact test of the 1997two-door Ford Thunderbird. To constructFigure 6. accelerometers critical locations-the locationsshown at in Figure 3 - were integrated to obtain the velocities. During the first 140 msec, all the velocities progressto a common velocity, vr, which is 32 kmph in the caseof the Ford Thunderbird. The velocity of the interior door goesfrom rest to 56 kmph in 11 msec. In the caseof the Ford Thunderbird, the velocity rises quite high, above the velocity of the striking barrier. The occupantis not moving at all during these initial 23 msec. As suggestedby Lau et al., the dummy is at rest and must receivequite a punch when the door intrudes inward about 132 mm.

L.-AU
\ I

I [ 1

5f

I/

I f

d.M

0.05

0.10 liinsecx4S

0.15

Figure 7. Velocity

of the MDB and Toyota Corolla.

To quantify the range in door velocities, consider Figure 8. Shown is a corridor for the door velocity observedin twenty-sevenNCAP side impact crash tests. The velocity of twenty-seven vehicles was computedat the mid B-pillar location of the struck vehicle. The two curves of the corridor are the plus one and minus one standard deviation curvefor the twenty-seven tests. For perspective, the averagevelocity of the center of gravity of the striking vehicle - the moving deformablebarrier - is drawn.

2494

50 +1 STD. DEVIATION 40

Integrating the velocity curve of the door. curve 6, in Figure 9, from t=Otot=&, will computethe distance. This distanceis also reportedas the arm-to-dooror hip-to-door distance. At some point in time, the occupant and the interior doorreacha commonvelocity. In Figure 5, this is markedby t,. By computingthe areabetween occupant the and door velocity curves from time b to time t,, one can determinethe amountof doorpaddingand structurecrush and occupant chest crush. Gabler et al. [4] define Door Effective Padding Thickness (DEPTH) as the relative displacement between doorand occupant the from the time of occupant-door contact until the time of occupant-door separation. Figure 9 below illustrates theseareas. From the crash observer perspective,DEPTH is the amount s which the occupanttorsodeformsplus the amountwhich s the occupant crushes the door padding and interior s structure.

Figure 8. Average Velocity Standard Deviation.

of Mid B Pillar with + 1

There are a possible choice of six door accelerometer locations:the left front sill, theleft front door centerline,the left front door mid-rear, the left front door upper centerline,the left lower B-post, and the left mid Bpost. The vehiclesmid-B pillar was chosenbecause, in general,this sensorperformedsatisfactory. Other sensors may have rotatedor had curvesthat did not approachthe fYinalvelocity. It is important to realize that for some vehicles, the peak velocity may have occurredat another accelerometer location. Generallyspeaking,the maximum velocity of the door varied between two peaks for the set of all door velocities observedim these laboratory tests. One is a maximum velocity that has an apogeearound the final velocity of the striking and struck vehicle. This is commonly termed the soft stroke of the impacting door. The secondtype of peak velocity occurs when the door velocity exceedsthe striking vehicle velocity. At this s time, apunch is said ito haveoccurred.
OCCUPANT CRUSH OF THE INTERIOR DOOR

t0

TIME

t1

Figure 9. Door and Occupant

Velocity

Curves.

Looking baclk at Figure 5, (or Figure 6 in the case of a specific car, the Ford Thunderbird),onesees door the s velocity, curve @, rise to a maximum. In part, the interior door beginsto decrease velocity because its occupantand door collide, and reaction forces are directed from the occupant,curve @), onto the door, curve 0. The initiation of this interaction is indicatedby time h in Figure 5. It is 17 msecin Figure 6. The distancebetween doorandthe occupant the is determined by computing the area between the door velocity curveandthevelocity curveof the occupant (which is zero at the mome:ntthe door contacts the occupant).

Figure 10 shows a plot of the Thoracic Trauma Index, TX, for the thirty-two cars versus the occupant crush of the interior door. The data is included in routinewas Appendix 1. of this paper. A linear regression compiled through the thirty-two data points and the Rvalue was computed 0.48.Equation l., below, describes to this relationship. These data suggest that there is a correlationbetweenTTI and the single variable, occupant crush of the interior door. [ 1l]

TTI = -0.146 DEPTH + 103.99

(1.)

2495

140 I

To determinethe pelvic lead. one calculatesthe difference in time betweenthe peak accelerationof the pelvis and the thorax (caused impact with the intruding by door), as shown in the equationbelow. Figure 11 shows two typical accelerationcurvesfor the pelvis and spine.
Pelvic Lead = t tono- t pe,V,S

-ii- 80 i2 E 60 40 20 0 0 100 200 300

(2.1

where.
t tOrrc time at maximum torso acceleration. = and

t PelV,S = time at maximum pelvic acceleration.

DEfTH [ml
Figure 10. DEPTH vs TTI.

VERTICAL INTERIOR

INTRUSION DOOR

PROFILE

OF THE

As Reference6 suggests, makessense,from a it biomechanicalpoint of view, to have the intruding door loadthe parts of the occupantbodythat canwithstand the s forces the most without trauma. Reference advisesthat 6 it is important to load the pelvic region during the initial occupant-door contact. As Hobbs explains, it is better to distributethe loadingalongthe entire torsoof the occupant, beginningfirst with the pelvis, than to havethe intruding structure only strike the shoulderand torso, where many important organsare located. [5] By referring to Figure 2, it can be seenthat the d!muny, SID, has a lateral accelerometer the pelvis. in There are two lateral accelerometers the upper and on lower ribs. The pelvis is a relatively rigid structurewhile the torso has a soft simulatedarm over a stiff rib cage. If the dummy is being loaded by an interior door that is vertically aligned, then the pelvis accelerometer should havean initial response the sametime or slightly before at the beginning of the rib cage acceleration. Should the pelvic signal start significantly beforethe signal at the rib cage,then the pelvis is contactingthe door first. This lead is commonlyreferredto as the pelvic lead. The pelvic lead phenomenonhas been applied to the design of some vehicles. [6] Basically. it amounts to the pelvis being impactedaheadof the thorax.

TIME

Figure 11. Occupant Thorax.

Response

of Pelvis and

For the thirty-two NCAP side impact tests,Figure 12 showsa plot of the ThoracicTraumaIndex, TTI, versus the pelvic lead. Thesetest data suggested the pelvic that leadwill introducebeneficialeffectsto the thoracicportion of the SID. In other words, more pelvic lead lessensthe severityof thoracic injury. Thesedata suggest thereis that a modestcorrelationbetweenTTI and the single variable. routine was compiledwith pelvic lead. A linear regression thirty-two datapoints and the R-valuewas computedto be 0.37, with the equationbelow, describingthe relationship.

[Ill

TTI = -0.509 Pelvic Lead + 95.05

(3.1

2496

EFFECT LEAD

OF COMBINING

DEPTH

AND PELVIC

In the Introduction sectionand the previoustwo sections,two important considerationsfor protecting an occupantin side impact were discussed. Using the data from thirty-two NCAP side impact tests, the dummy s response examinedfirst as a fimctlon of DEPTH and was then as a function of pelvic lead. Alone, neither of these variablessatisfactorilyexplainsthe response the dummy of in thesetests. The next step in this study is to investigate if the two variablestogetherincreaseour ability to explain the dummy response. s A linear regression analysiswas performedusing a routine to determine the regressioncoeffkients and confidence levels. [ 1I] Two independentparameters, DEPTH and pelvic lead, were chosenfor the regression analysis. A relationship betweenthese two independent variables and the Thoracic Trauma Index, TTI, was computed. The R-value for the combinationof variables, DEPTH and pelvic lead, is 0.60 with the relationship described Equation 4. [l l] Figure 14 graphsthe TTI, by recorded the NCAP tests,versusthe linear combination in of DEPTH and pelvic lead. Figure 14 illustrates that TTI increases, monotonically, as the linear combination of DEPTH and pelvic lead increases. This linear model explainsabouta third of the variability in the data.

-2 -1 0

1 2

9 IO 11 12 13 14

t(ton;o) - t(pelvis)

[msec]

Figure 12. Pelvic

Lead vs TTI.

Figure 13 plots the Pelvic Lead vs the maximum pelvic acceleration value. As shown,the Pelvic Leadand maximum pelvic acceleration not necessarily are relatedto each other. In other words, greaterPelvic Lead may not necessarily developlarge pelvic acceleration.

TTI = -1.42 Pelvic Lead - 0.14 DEPTH + 112.5

(4.)

.inn

U'IE

Pelvic Acceleration

[G s]

Figure 13. Pelvic Lead vs Pelvic Acceleration.

2497

90 E
80

60

FIGURE 15. Thoracic (Reference 12).

70 80 90 100 110

Trauma Index Risk Function

-1.421 Pelvic Lead - 0.141 DEPTH + 112.47

From the probability values, a star rating for an occupantwas developed.The following levelsare usedto designate stars: the
QQQQQ =

Figure 14. Linear Combination DEPTH.

of Pelvic Lead and

5% or less chanceof seriousthoracic and upperabdominalinjury 6% to 10% chanceof seriousinjury 11% to 20% chanceof seriousinjury 21% to 25% chanceof seriousinjury 26% or greaterchanceof seriousinjury

To understand the two independent how variables interact to mechanicallyaffect the dummy response, s the datain Figure 13 may be plotted as DEPTH versuspelvic lead. A datapoint would then be the value of TTI recorded in the NCAP crashtest. For this type of crossplot, greater clarity is obtainedby arrangingthe TTI valuesinto groups and giving each group a nemonic symbol. On the cross plot, a data point would then appearunder the symbol of the group to which its value belongs To plot DEPTH versuspelvic lead, the TTI value was first convertedto a star rating. For thosereaders familiar with the star rating not methodology,the side star rating systemis basedon the thoracic injury function curve developed the Thoracic for Trauma Index. This thoracic injury function curve is containedin the final regulatoryevaluationfor FMVSS No. 214 [12] and is shown in Figure 15. This function relatesthe probability of an AIS zz4 thoracic and upper abdominalinjury to TTI in a lateral impact.

QQQQ QQQ as Q

= = =

Using the risk curve, the star ratings correspond a to rangeof TTI values.
QQQAQ QQAQ QQQ aa Q = = = = zr

TTI < 57 57 < TTI 2 72 72<TTI 191

91 <TTI s.98 TTI > 98

In short, the star rating methodologyconverts the continuousTTI into a categoricalvariable. Appendix 2. provides star ratings for forty-six NCAP side impact vehicles. Figure 16 shows the test data on a graph of DEPTH versus pelvic lead. For each of the thirty-two

2498

NCAP tests,the dummy response represented its star s is by rating. Superiordescriptors,of the mechanicalinput into the dummy, should partition different star ratings further apart with minimal overlaps. For poorer descriptors,the star ratings should be intermixed without clear separation. In Figure 16, it can be:seenthat DEPTH and pelvic lead begin to separatethe outcome variable, representingthe dummy response. Those dummy responseswith three s stars are grouped together. The four star ratings are generally higher than the three star ratings but are mixed with the three star ratings. Figure 16 suggeststhat for a given value of crushing door padding, a lower thoracic response may be obtainedby impacting the pelvis roughly 10 ms before the torso.
Test Data as a Function of DEPTH and Pelvic Lead

door in the struck vehicle was calculated. It was found that the speedsof the thirty-two intruding doors appearedto vary over a wide range. Somedoorshad a maximum speed that reached only as high as the velocity of the struck vehicle. Other intruding doors reacheda speedroughly twice the final velocity of the struck vehicle. The Door Effective PaddingThickness (DEPTH) is the relative displacement betweenthe door and occupant from the time of occupant-doorcontact until the time of occupant-door separation.The DEPTH was calculatedfor thirty-two cars crashed in NCAP side impact. The correlationbetweenthe dummy response the DEPTH s and was modestfor this data set. The side impact dummy hasaccelerometers the in torso and an accelerometer the pelvis. In a side impact, in a pelvic signal starting significantly beforethe torso would indicate a pelvic lead and mean the door is contacting the pelvis before the torso. The pelvic lead is defined as the time of the pelvic response subtractedfrom the time of the thoracic response. The pelvic lead was calculatedfor the thirty-two cars crashed in NCAP side impact. A weak correlation betweenthe dummy response the pelvic s and lead was found for this data set. No single variable fully explains the responseof the dummy during a side impact event. The linear combination of DEPTH and pelvic lead accountfor about a third of the variation of this data set. The next step in this study will be to investigate other variables with the objectiveof more completelydescribingthe intensity of the mechanicalinput and the responseof the occupantin the struck vehicle.

Pelvic Lead [msec]

Figure 16. Linear Combination DEPTH at a Constant TTI. CONCLUSIONS

of Pelvic Lead and

The USA NCAP conducted thirty-two side impact crash tests. These tests were based on the testing methodologyof FMVSS No. 214. The deformablemoving barrier was traveling at 61.6 kmph just before hitting the struck vehicle. Accelerometers were installed in a variety of locations about the door panel, A-pillar, and B-pillar of the struck vehicle. In previousside impact testing, struck dooraccelerometers frquently exhibitedanomalies because they were subjectedto severeimpact loading. An analysis was conductedof the accelerometers usedin the thirty-two NCAP tests. It was determined that many of the accelerometerssurvived the side impact and produced satisfactorysignals. Using the accelerometers the interior door, Ain pillar, and B-pillar, th.e maximum velocity of the interior

2499

REFERENCES TrafJc Safety Facts 1996, National Highway 1. Traffic Safety Administration, U.S. Department of Transportation, ReportNo. DOT HS 808 649, pp. 71 and 73, December1995.

North Carolina, 1991.

Final Regulatoq Impact Analysis: New 12. Requirementsfor Passenger Cars to Meet a Dynamic Side Impact Test FWSS 2 I4, National Highway Traffic Safety

Administration, Report No. DOT HS 807 641, August 1990.

2.

NHSA Releases Side Crash Test Results irz New Consumer Information Program, NationalHighwayTrafi%

SafetyAdministration,U. S.Department Transportation, of PressRelease NHTSA 21-97, 11 April 1997. No.

Euro NCAP Crash Tests, UK Department of 3. Transportand What Car?, January1997.

Gabler, H.C., Hackney, J.R., Hollowell, W., 4. DEPTH: A RelationshipBetween Side Impact Thoracic Injury and Vehicle Design, 12th InternationalTechnical Conference ExperimentalSafetyVehicles, Gorteborg, on Sweden1989. Hobbs,CA., The influenceof Car Structures and 5. Padding on Side Impact Injuries, 12th International Technical Conference Experimental Safety Vehicles, on Cnrteborg,Sweden1998. Saab9-5 is packed with innovations,Automofive 6. Engineering, The Society of Automotive Engineers,400 Commonwealth Drive, Warrendale,Pennsylvania15086, September 1997. Lau, I. V., Capp,J. P., and Obermeyer, A., A J. 7 Comparison Frontal and SideImpact: CrashDynamics, of Countermeasures Subsystem and Tests,Thirty-fifth Stapp Car Crash Conference Proceedings, November1991.
8. Strother, C. E., and Morgan, R. M., A fundamental look at side impact, Proceedingsof the American Association AutomotiveMedicine,Louisville. for Kentucky, October1979.

Rao, A., Thompson G.J., Mucino, V.H., and 9. Smith, J.E., Crash Analysis Response a Mid-size Car of Stibjected SideImpact,SAE InternationaICongress to & E::position,Detroit, Michigan, February1997. B&dell, D., Stephens,G., and Meissner U., 10. Collision Performanceof Automotive Door Systems, SAE PublicationNo. SP-1045.1994. Littell R. C., RudolfF. J., and Spector C. SAS P. 11. System LinearModels,31d SAS Institute Inc.. for Ed., CG

2500

Appendix 1. Test Data for Thirty-two Test No. Make & Model

NCAP Side Impact Tests Model Year Vehicle Test Wt. ka TTI a&I LUR Depth mm Pelvic Lead msec

3 2 l 1 4

I-1 9

ChevroletCamaro2-dr Cadillac Deville 4-dr Cavalier2-dr ChevroletLumina 4-dr ChevroletMalibu 4-dr

Ford Crown Victoria 4-dr

10 Ford Escort 4-dr 11 Ford Taurus4-dr 12 Ford Thunderbird2-dr 14 HondaCivic 4-dr

17 Mazda 626 4-dr

19 Nissan Maxima 4-dr 20 Pontiac {Grand 4-dr AM SubaruLegacvAWD 4-dr

1 26 Volvo 850 4-dr

27 Buick Century 29 ChevroletCavalier4DR 1 30 Ford Escort ZX2 31 Ford Mustang 32 Mercedes Benz C-230

1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1998 1998 1998 1998 1997 1998

2085 1450 1744 1758 1618 1741 1538 1506 1995 1328 1756 1814 1470 1241 1543 1305 1424 1496 1618 1581 1267 1562
1601

57.8 123 86 61 100 72 75.8 73 68.6 88 74 91 96 83.1 102 93 92.9 84 70 109 88.2 88.4
82.2

159 71 99 168 94 109 141 88 186 108 171 132 141 72 114 84 118 249 196 144 110 192
127

5 9 7 4 0 13 8 6 6 9 11 7 13 8 6 -1 2 11 2 9 3
6

1312 1149
1723 1766

89 83.3
62 82

99 81 192
206

8 10 11
6

1805 1592 1344 1601

1626

79 105 100 87
86

110
73 66

6
3 5

69
1103

5
2

2501

Appendix 2. Star Ratings for Forty-six

NCgP Side Impact Tests

2
3 4

5 6 7
8

9 IO
11 12 13 14 t -a 6 17 15

ChevroletCamaro2-dr ChevroletCavalier 2-dr ChevroletLumina 4-dr ChevroletMalibu 4-dr DodgeIntrepid 4-dr Dodge Stratus4-dr Ford Contour4-dr Ford Crown Victoria 4-dr Ford Escort 4-dr Ford Taurus4-dr Ford Thunderbird2-dr Honda Accord 4-dr Honda Civic 4-dr Mazda 626 Kia Sephia4-dr Hyundai Sonata4-dr Mitsubishi Galant4-dr NissanMaxima 4-dr Pontiac GrandAM 4-dr SaturnSL 4-dr SubaruLegacyAWD 4-dr Toyota Canuy 4-dr Toyota Corolla 4-dr ToyotaTercel2-dr Volvo 850 4-dr Buick Century Buick LeSabre ChevroletCavalier4DR Ford Escort ZX2 Ford Mustann Mercedes Bek C-230 _ Mazda 626 DodgeNeon PontiacBonneville Honda Civic 2DR Nissan Altima ToyotaAvalon

1997 1997 1997 1997

1997

1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1997 1998 199s 199s 1998 1997 1998 1998 1998 1998 1998 1998 1998

83.3 107.5 54.4 82.9 78.3 74.0 63.9 69.3 78.0 71.0 71.5 94.4 78.5 113.9 70.7 79.0 70.0 65.9 104.0 86.0 84.5 76.4 81.8 86.7 57.4 76.0 63.3 99.2 85.9 76.5 67.1 56.7 93.7 74.4 85.1 81.5 47.6

81.9 135.8 61.1 95.2 80.4 75.8 69.5 65.5 99.0 78.0 87.7 78.0 86.9 109.1 90.4 90.9 74.0 63.3 94.5 83.9 81.2 82.0 75.3 91.6 50.5 77.0 64.8 74.4 99.2 80.4 80.8 67.5 87.0 72.9 107.9 89.8 51.4

88.4 110.2 61.4 105.6 61 4.5 75.8 76.4 67.8 77.0 70.0 93.4 97.4 79.3 95.4 95.1 89.7 94.0 73.7 114.4 90.4 92.2 82.4 96.4 74.9 66.5 86.8 93.2 110.7 100.9 94.3 91.0 93.3 94.0 94.1 78.9 87.1 61.1

83.0 115.8 69.5 109.8 108.9 108.7 82.0 80.9 120.0 97.0 131.8 117.0 107.2 114.5 112.8 84.7 98.0 95.3 102.2 123.8 130.8 88.6 106.9 79.2 70.5 128.8 98.1 140.7 119.8 90.1 79.5 135.7 102.0 111.5 92.1 93.0 106.3

aaa A c&a* a *a** **72 fri? r* T&x?* **a

a** **a aa ak2 aa a B&B **** a a** *lx? *a* a** *** **e* a** **a B a a** *cr* *a* a* aaa aa **a CrrzrzrQCr

-cr*** crcr a** a** a** aa

aaaa 1

*zr*rzr *a*
**A a ABA aaa a*** aa cl* a** -ix& a** ND **a **a aaaa ND aCr* alla aacr aaaa aaa **** a** crcrcr aa crcrcr aa* Adaa I

18 19
20 21 22

23
24 25 26

27
28

29
30 31 32 34 35

;3 ii

_36
37 -38

2502